thanushshetty9353/Federal-SLM-System — reverse-engineered prompt

Build me a privacy preserving document intelligence platform for sensitive files like medical records, legal documents, insurance claims, research data, and financial reports.

I want organizations to be able to log in, upload scanned PDFs or images, run OCR, and have a local Small Language Model turn the text into clean structured JSON insights. The raw documents should stay with the organization. Add a dashboard with separate roles for admin, organization users, and researchers. Admins should approve users, manage schemas, and view audit logs. Organization users should upload documents, manage datasets, and start training. Researchers should view analytics and test insights.

Include federated learning so multiple organizations can improve a shared model without sharing their raw data. Add blockchain style audit logging for important actions like uploads, OCR processing, model updates, and training runs. Make the UI clean with charts, status screens, upload flows, and clear error messages.

Use the stack described in the repo, and look up current docs online if needed. Make it runnable locally and with Docker.